On native and contact-induced grammaticalization_

Document Sample
On native and contact-induced grammaticalization_ Powered By Docstoc
					On native and contact-induced grammaticalization: The case of Malay empunya
Chinese University of Hong Kong†

Foong Ha Yap

Abstract Lexical items without any inherent attitudinal meaning sometimes evolve into stance markers. This paper examines one such phenomenon, and addresses the following questions: How do such stance markers emerge? What language-internal forces facilitate their development? In cases involving language contact, how do language-external forces interact with language-internal facilitations? This paper examines both diachronic evidence from classical Malay texts (17th to 19th century) and the influence of contact languages (in particular southern Chinese dialects) on the grammaticalization of Malay lexical noun empunya into a stance marker in sentence-final position, and also as an attitudinal intensifier in pre-adjectival position. Converging evidence points to the pivotal role of pronominal uses of empunya in facilitating the development of new grammatical and pragmatic functions. This paper further identifies similar phenomena in other Asian languages.

1. Introduction The Malay morpheme empunya, originally a noun meaning ‗master‘ or ‗owner‘, has undergone significant phonological reduction and grammatical abstraction over the past few centuries, giving rise to the forms punya and mya in some Malay varieties,1 at the same time adding to the speaker‘s repertoire a new range of possessive morphemes and speaker mood or stance markers. This paper provides a diachronic of how this development was facilitated, highlighting in particular the interaction between native and contact-induced grammaticalization. We organize this paper as follows. §2 provides a brief description of the Malay language. §3 then identifies the major grammatical categories and functions of (em)punya, while §4 traces the pathways of change for empunya. More specifically, §4.1 examines how lexical noun empunya evolved lexical verb uses, while §4.2 and §4.3 trace the grammatical development of (em)punya into genitive and pronominal forms respectively; §4.4 then traces the link between pronominal punya and stance-marking punya/mya. §5 focuses on the influence of language contact on the emergence of grammatical uses of punya. §6 compares the grammaticalization paths of empunya with those of stance markers from other neighboring Asian languages that are known to have evolved via the determiner pathway. This provides us with a means of identifying robust pathways (if any) for the rise of stance markers across languages. 2. Background on the Malay language


Malay is a Western Austronesian language that has long served as a lingua franca among the islands of Indonesia and throughout neighboring Brunei, Malaysia and Singapore. The Malay variety discussed here is spoken along the western coast of peninsular Malaysia, and the classical texts analyzed are representative of the speech varieties spoken/written in the vicinity of the Straits of Malacca from the 17th to 19th century.2 Malay is an SVO language in terms of canonical word order. There is productive prefixing (meN-, ber-, ter-, di-), suffixing (-kan, -i), and circumfixing (meN-...-kan/-i, di-...kan/-i) on the verbs to mark a variety of features including voice (active, passive and middle), volitionality (e.g. intentional vs. accidental), quantification (e.g. multiplicity) and aspect (e.g. iterativity), among others, though in colloquial speech the use of bare verbs is also very common. Nominalizing affixes are also very productive, with ke-...-an, peN-...-an and transforming verbs into nouns. Enclitic –nya occurs with high frequency, and serves a wide range of functions, including third person pronoun, possessive marker, definiteness marker, nominalizer, adverbial marker, and epistemic marker (see Englebretson 2003 for a detailed discussion of -nya in colloquial Indonesian). Malay also has another highly versatile nominalizer, namely yang, which is also identified with a wide range of functions, including relative clause marker, complementizer, topic marker and definiteness marker (see van Minde 2003 for a recent discussion). Structurally, except for its referential use as third person pronoun (e.g. Jangan di-usik-nya ‗Don‘t touch (disturb, or provoke) him/her/it‘), enclitic – nya is essentially head-final, much like pronominal and stance marker punya/mya (see §4.3 and §4.4), whereas yang (whose lexical origin is unknown) is head-initial (see §5.1). Malay makes extensive use of postnominal modification—this applies to possessive and relative clause constructions, as illustrated in (1) and (2) below. The canonical structure, essentially, is [Nhead noun] [XP], where XP is the post-modifying phrase. (1) Dia guna [kereta] [orang lain]. 3SG use car person other ‗S/he used someone else‘s car.‘ (2) Dia suka [gambari ] [(yang) saya lukis ___i] (itu). 3SG like picture REL 1SG draw DEF ‗S/he likes the picture that I drew.‘

In terms of linear order, as seen in (1) above, canonical possessive constructions involve a juxtaposition of possessee and possessor noun phrases, more specifically [possessee NP] [possessor NP], without recourse to an overt genitive marker or linker. In relative clause constructions, the morpheme yang often appears optionally between the head noun and the modifying clause. Arguably, yang functions either as a relative pronoun or as a head-initial complementizer introducing the complement clause that modifies the head noun, namely gambar (‗picture‘) in (2) above. From a grammaticalization perspective, it is plausible that a relative pronoun has further become reanalyzed as a complementizer. Adelaar (1985) has suggested that yang could have been derived from a third person pronoun ia plus a linking particle –N; this account is not unlike English demonstrative pronoun that evolving into a relative clause marker (Hopper & Traugott 2003). Factive complementations in Malay are post-verbal, often introduced by complementizer yang, much like English that-clause constructions. As illustrated in (3), the complement clause—in this case, (yang) saya akan pindah ‗(that) I will move (to a new place)‘— immediately follows the matrix verb, tahu ‗know‘ in this particular instance. Worth noting is

the optionality of complementizer yang, which suggests that noun complement clauses in Malay need not be accompanied by an overt complementizer. (3) Aku tahu [(yang) dia akan pindah]. 1SG know COMP 3SG FUT move ‗I know (that) s/he will move (to a new place).‘ Another important point to emphasize, for the purpose of this paper, is that the canonical relative clause and factive complementation constructions are head-initial constructions. This head-initial configuration is often overtly marked by functional head yang, as noted in (2) and (3) above. In contrast, when used as relative clause marker and stance marker, (em)punya appears in head-final position (see §3.3 and §3.4). In subsequent sections we will trace the rise of head-final empunya, focusing on evidence of language-internal structural facilitation (see §4), as well as language-external influence in the form of extensive contact with other languages, in particular southern Chinese dialects (see §5). 3. Grammatical categories and functions of (em)punya In this section we examine the range of functions of the fairly versatile morpheme empunya. Analyses of classical Malay texts and contemporary colloquial usage reveal that (em)punya can be used lexically as a noun or a verb, grammatically as a genitive and possessive pronominal, and pragmatically as an epistemic, attitudinal and interactional marker (which we here refer to as speaker mood or stance marker). We briefly examine each of these functions below. 3.1 Lexical noun empunya The etymology of empunya can be traced back to a lexical noun empu. According to dictionary sources such as Kamus Dewan, the noun empu is synonymous with another noun tuan, and both convey the meaning ‗master‘ or ‗master craftsman‘. Empunya is also equated with tuannya ‗master‘ and pemiliknya ‗owner/possessor‘, with third person enclitic –nya often yielding either a pronominal possessive interpretation (e.g. ‗his/her/its master‘) or a more general definiteness reading (e.g. ‗the master‘), as highlighted in (4).3 (4) empu -nya master 3SG ‗(his/her/its/the) master/owner/possessor‘ Lexical noun uses of empunya with the meaning of ‗master‘ or ‗owner‘ are attested in numerous classical texts at least as early as the 17th century, as shown in (5). (5) Siapa-siapa baik empunya kuambil whoever regardless owner 1SG-take karena ia datang ke hadapan rumahku kuambil. because 3SG come to front house-1SG 1SG-take ‗No matter who the owner, I will take (it=the deer); because it came up right to my house, I (will) take it.‘

(SR 746:6, original text 16th century; available ms. 17th century)


While lexical noun empunya could occur independently as in (5), it could also be combined with another noun as in (6). This N-N type construction provides more specific information about the ‗master‘ or ‗owner‘, through association with the type of craft engaged in, or the particular object being possessed, or the domain of possession. As shown in (6a-c) below, examples include empunya ce(ri)tera ‗a master of stories, i.e. a storyteller‘, empunya kijang ‗possessor or owner of a deer‘, and empunya negeri ‗possessor or sovereign head of a kingdom, i.e. ruler or king‘. (6) a. empunya ce(ri)tera ‗the storyteller‘ (IP 82:48, 1600) b. empunya kijang ‗the owner of the deer‘ (SR 752:3, 16th/17th c.) c. empunya negeri ‗the ruler of the country‘ (SR 314:8, 16th/17th c.) In this N-N type construction, empunya is the head noun, while the second noun modifies it. This constitutes a head-initial construction, not unlike the [Nhead noun] [XPmodifying phrase] structure identified earlier for canonical possessive constructions. It differs however from the head-final constructions that characterize empunya-type relative clause and speaker stance constructions (see §3.3 and §3.4 respectively). 3.2 Lexical verb empunya Use of empunya as a lexical verb was also already evident in 17th century texts such as Hikayat Inderaputera. As shown in (7), lexical verb empunya can convey a sense of ‗possessing something or someone (the latter as in kinship relations)‘—in this case, empunya cucu ‗possess(ing) a grandchild‘. In such examples, a lexical noun reading of empunya (e.g. in the sense of ‗owner/possessor of a grandchild‘) would be rather awkward. (7) Sekarang tiada aku empunya cucu now NEG 1SG have grandchild karena cucuku sudah dibuangkan ke laut. because grandchild.1SG PERF PASS.throw.TRANS to sea ‗Now I no longer have a grandchild
(? Now I am no longer the possessor of a grandchild) ,

since my grandchild has been hurled out into the sea.‘

(IP 172:36, available ms. 1600)

It is worth noting that lexical noun uses of empunya were often accompanied by definiteness marker yang, as shown in (8) below, while lexical verb uses of empunya came to be increasingly marked by verbal affixes meN-...-i, as in (9). (8) serta ... minta segera dipertemukan kepada and request immediately PASS.introduce to yang empunya delima itu DEF owner pomegranate that ‗and ... (he) asked immediately to be introduced to the owner of the pomegranates‘ (IA 96:29, original text ca 1650, ms. 1775) (9) lagipun tiada ia mem-punya-i ilmu


moreover NEG 3SG PREF-possess-SUF knowledge ‗moreover, he does not possess any knowledge‘ (PAK 15:33, 19th century) 3.3 Genitive and possessive pronominal empunya (Em)punya is also used as a grammatical possessive marker, as in (10). This type of possessive construction, which involves a [possessor NP + (em)punya + possessee NP] construction, was especially productive in 19th century texts, such as Misa Melayu and Pelayaran Abdullah ke Kelantan, but is now highly restricted to colloquial usage. (10) disuruh kerah masing-masing punya jawatan PASS:tell strengthen GEN position ‗each was ordered to strengthen their position‘ (MM 43:30, 19th century) As shown in (11), use of (em)punya as a possessive pronominal was also frequently attested from the 19th century onward. Crucially, in such constructions (em)punya identifies a possessee in relation to its possessor (the genitive function), while at the same time alluding to the morphologically unrealized possessee as well (the pronominal function). Consequently, possessive pronominal (em)punya allows us to focus on the possessor, while still referring to the possessee. This is particularly useful in contexts such as (11), where ‗possessor identity‘ is being contested.4 (11) Kata seorang, ―Ini aku punya‖; say one.person this 1SG GEN kata seorang, ―Aku punya.‖ say one.person 1SG GEN ‗Said one person, ―This is mine‖; said another, ―(It‘s) mine.‖ ‘
(PAK 111:8, 19th century)

Worth noting here is that, whereas (em)punya possessive constructions can be used pronominally (i.e. without an overt possessee NP as head noun), this option is not available to the non-(em)punya (i.e. canonical) possessive constructions. This difference is highlighted in (12) and (13) below. (12) [possessor NP + punya + possessee NP] ---> [possessor NP + punya + Ø] saudagar itu punya emas saudagar itu punya merchant DEF GEN gold merchant DEF GEN ‗the merchant’s gold‘ ‗the merchant’s‘ (13)
[possessee NP + possessor NP]


[*Ø + possessor NP]

emas saudagar itu gold merchant DEF ‗the gold of the merchant‘ 3.4 Stance (*em)punya

saudagar itu merchant DEF ‗the merchant/*the merchant’s‘


In the classical texts examined, there were no tokens of stance empunya. However, use of its phonologically reduced forms (i.e. punya and mya) to signal certainty and confidence in one‘s own utterance (sometimes in anticipation of potential skepticism on the part of the addressee) is fairly common in contemporary colloquial Malay. Examples of such assertive use are shown in (14) and (15). (14) Dia tak akan datang punya/mya. 3SG NEG FUT come STANCE ‗S/he won‘t be coming (I can assure you).‘ (15) Yang ini (pasti) masuk punya/mya! DEF this surely enter STANCE ‗This one will (surely) go in!‘

(see also Gil 1999)

Without the use of punya/mya, the utterance in (14) would simply be a factual statement. The addition of sentence-final punya/mya, however, makes it a ‗trust me‘ or ‗I‘m telling you‘-type assertion. In (15), the presence of punya/mya amplifies the display of confidence on the part of the speaker. Punya can also function as an intensifier in pre-adjectival position, to express assertions that are often laced with strong feelings, including feelings of awe, as in (16), or feelings of incredulity or even annoyance, as in (17). (16) Punya halus pasir di pantai ini. STANCE fine sand LOC beach this ‗So fine, the sand on this beach.‘ (17) bodoh aku ‘ni! stupid 1SG this ‗So stupid of me!‘


Pre-adjectival uses of stance punya appear to have evolved as an extension of its genitive and associative linking functions, in particular via scalar-type [begitu + punya + Adjective] constructions, as outlined through the three stages in (18a-c) below. (18) a. Dia ‗ni begitu punya bodoh! 3SG DEF like.that LNK stupid ‗S/he was so stupid!‘ (< ‗S/he was that stupid!‘) b. bodoh! 3SG DEF STANCE stupid ‗S/he was so stupid!‘ Punya bodoh dia ‘ni! STANCE stupid 3SG DEF ‗So stupid s/he was!‘ or ‗(It was) so stupid of her/him!‘ Dia ‗ni punya


In (18a), punya links the deictic adverb begitu (‗like that‘) to the adjective bodoh (‗stupid‘), yielding an interpretation in which the adverb functions as an intensifier for the adjective. With emotions running high, intense prosody spreads all the way from deictic adverb begitu to adjective bodoh, infusing linker punya with emotional overtones as well.5 As seen in (18b), linker punya can also function as an intensifier in the absence of the deictic

adverb begitu. Such reanalysis is, in some ways, not unlike the extension of negative polarity meaning from ne to pas in French. In both cases, collocational frequency plays a facilitative role. In (18c), we see evidence of greater syntactic freedom—with the ‗intensifier punya + adjective‘ construction now preposable to sentence-initial position. At any rate, punya bodoh ‗so-ooo stupid‘ clearly is intensely ‗stance-ful‘. In this paper, we treat this class of attitudefilled pre-adjectival punya constructions as a subset of stance punya constructions. Gil (1999) also provides examples of post-adjectival punya constructions such as Besar punya! (‗It‘s so huge!‘). This intensifying usage of punya appears to be linked to associative uses of the [XP] punya [NP] type, where [XP] denotes a prenominal modifying construction, of either adjectival or predicational status. As highlighted in (19) below, elision of the head noun budak ‗child‘ triggers reanalysis of punya from associative marker to stance marker. (19) a. (Begitu) bodoh punya budak! like.that stupid LNK child ‗Such a stupid child!‘ b. (Begitu) bodoh like.that stupid ‗So stupid!‘ punya!

Note that stance uses of punya in pre-adjectival and post-adjectival position are both of the intensifying type, diachronically linked to a deictic interpretation via the similative adverb begitu ‗like that‘. The earliest (and sole token) of a begitu punya + Adjective construction in the Malay Concordance Project database was found in Hikayat Abdullah (1843 AD), and is reproduced below: (20) Maka kata Tuan Morgan, ―Di mana lagi Tuan boleh dapat begitu punya untung, kerana ia itu kepala doctor, ia datang ke mari mengambil-ambil angin, seribu doktor di sebelah sini tiada boleh sama dengan dia, ia general doctor. …‖ Then Mr. Morgan said, ―Where else could you get so lucky (< that lucky), because he is the head doctor, he comes here to relax (lit. ‗get some air‘), a thousand doctors over here cannot compare to him, he‘s the general doctor.…‖
(AM, 215:36; 19th century text)

While there may still be a problem of lacunae (i.e. insufficient texts to adequately represent the historical development of empunya), as well as a problem of register (i.e. punya is more likely to be found in colloquial speech), available textual evidence suggests that preadjectival stance punya constructions appeared earlier than sentence-final stance punya constructions (see Table 1 in §4). As further highlighted in (21) below, sentence-final stance punya constructions are structurally more similar to post-adjectival (than to pre-adjectival) intensifier punya. An important difference between the two, however, is that the preadjectival type is an intensifying type, commonly used in exclamatives, particularly when the lexical noun referent is omitted, while the clausal sentence-final type conveys speaker‘s assertive mood with respect to the entire proposition. The latter thus involves a de-emphasis (and absence) of the lexical head noun (N), leaving super-linker punya in head-final position.
(21) a. pre-adjectival intensifier punya: b. post-adjectival intensifier punya: c. sentence-final stance punya: (begitu) + punya + AdjP (begitu) + AdjP + punya + (N) IP/VP + punya + (*N)


In the next section, we will examine in greater detail how (em)punya evolved from lexical to grammatical uses, with special emphasis on diachronic development as well as structural reanalyses. 4. Pathways of change for empunya Diachronic evidence from classical texts made electronically accessible via the Malay Concordance Project database (courtesy of Ian Proudfoot at the Australian National University) confirms that lexical uses of empunya (as nouns and verbs) were already attested in pre-17th century texts, while grammatical uses such as genitive and pronominal functions emerged later, becoming productive from the 19th century onward (see Table 1 below; adapted from Yap, Matthews & Horie 2004). Stance marking functions associated with punya and mya appear to be a more recent development, and are greatly restricted to the colloquial register. The development of empunya appears to have involved at least three major stages.
Table 1. Summary of diachronic development of (em)punya (adapted from Yap, Matthews & Horie 2004) Empunya
Noun 14th c. 16th c. 17th c. 18th c. 19th c. 9 29 66 137 37 N/V 9 2 7 3 3 Verb 6 5 28 13 16

Noun N/V Verb

Punya, mya
Genitive Pronominal

Punya, mya

11 12 10


11 4 62

1 87 1 18 1 pre-adjectival punya

-------------------------------------------------------------------------------------------20th c. Archaic usage of empunya Frequent use of definiteness Productive in colloquial Productive use of (e.g. frozen N-N forms marker yang for lexical noun usage. pre-adjectival punya such as empunya kerajaan). empunya and obligatory use of and sentence-final verbal affixes for lexical verb punya/mya in empunya, yielding the yang colloquial speech. (em)punya vs. mempunyai distinction.

(Note: Frequency counts for the 14th through 19th centuries are based on analyses of classical texts from the Malay Concordance Project, excluding Warisan Warkah Melayu, since this collection is multi-dated. We focused our analyses on prose, rather than poetry, collections.)

The first stage involves lexical category expansion, where empunya came to be used not only as a lexical noun but also increasingly as a lexical verb. It is worth recalling here that increasing use of verbal affixation makes the task of disambiguating between lexical nouns and lexical verbs much easier, and such affixation is now obligatory in modern Malay, yielding the frozen form mempunyai, meaning ‗possess, own, have‘. The second stage involves the emergence of grammatical uses of (em)punya, more specifically as genitive marker and pronominal morpheme. The third stage involves the extended use of the phonologically reduced variants punya and mya as stance particles. Note that stance punya appears in both sentence-final and pre-adjectival positions, while stance mya appears only in sentence-final position. This, in large part, is due to the intensifying function of attitudinal punya in pre-adjectival position, hence resistance to the highly phonetically reduced mya. Figure 1 highlights the three major waves of development for empunya.


Figure 1. The three major waves of development for empunya in classical and modern Malay 1st wave: lexical category expansion 2nd wave: syntactic restructuring 3rd wave: reanalysis of sentential nominalizers as mood markers LEXICAL NOUN empunya > yang empunya LEXICAL VERB empunya > mempunyai GENITIVE (em)punya & PRONOMINAL (em)punya STANCE (*em)punya/mya

In the rest of this section, we will examine language-internal facilitations in terms of structural viability. 4.1 From lexical noun to lexical verb Given that etymological evidence points to a lexical noun origin for empunya (see §3.1), it is reasonable to deduce that lexical verb usage of empunya is derived from a reanalysis of the lexical noun. Such reanalysis can easily be facilitated by zero-copula constructions, which appear with high frequency in Malay. Consider (22) below, which involves an equative construction with zero-copula. Overtly, the subject aku ‗I‘ is immediately followed by the noun complement empunya kajang itu ‗the owner of the deer‘. In the absence of a copula or linking morpheme, this noun complement can be reanalyzed as a verb predicate, yielding the interpretation ‗own (i.e. possess) the deer‘. (22) Aku empunya kajang itu 1SG own(er) deer that a. ‗I (am) the owner of the deer‘ b. ‗I own the deer‘ (SR 752:3, 16th/17th c.) Worth noting is that both syntactic and prosodic cues converge to yield the proper interpretation. Thus in (22a) a lexical noun interpretation of empunya will involve a distinct prosodic juncture (i.e. a brief but salient pause) between the subject noun aku ‗I‘ and the noun complement empunya kajang itu ‗the owner of the deer‘, arguably signaling the presence of a zero-copula. On the other hand, prosodic discontinuity preceding empunya is less clearly discernible in (22b). This is consistent with the interpretation that no zero-copula is intended here, and the position of the lexical verb is being assigned instead to a denominalized (i.e. verbal) empunya. 4.2 Emergence of genitive (em)punya Although already attested in 17th century texts, genitive punya did not become productive in literary texts till the mid-19th century. One possible reason for its late development is competition from an already well-established canonical possessive construction (see §2). Strong competition notwithstanding, reanalysis of lexical noun punya into genitive punya was structurally viable. Consider (23) below, where we see a cleft construction from 19th century Hikayat Bayan Budiman, with empunya interpretable either as a lexical noun (meaning ‗possessor‘), a lexical verb (meaning ‗to possess‘), or a genitive morpheme (like


English enclitic ‘s). Such indeterminateness (i.e. underspecification) could easily pave the way for the emergence of more abstract grammatical categories. (23) Bukannya hamba empunya kerajaan NEG.3SG 1SG.HUMBLE own(er)/GEN kingdom a. ‗It‘s not that I am the sovereign/ruler of this kingdom.‘ b. ‗It‘s not that I possess this kingdom.‘ c. ‗It‘s not my kingdom.‘
Maka sahut Tuan Puteri itu, ‗Sebenarnyalah kata tuan hamba. Adinda pun mengiringkan kakanda juga sama-sama; jikalau di negeri ini pun, bukannya hamba empunya kerajaan, karena sudah terserah kepada saudara hamba yang laki-laki itu.‘ (BB 165:6, original text 1371; available ms. 1852) ‗Then the princess replied, ‗Let it be as you say. I will go with you and accompany you, for even if I remain in this country, it’s not that I am ruler of this kingdom (or: it’s not that I possess this kingdom, or: it’s not my kingdom), for it has been handed over to male relatives of mine.‘

With respect to (23a) and (23b), we can see a very close link between lexical noun and lexical verb empunya. As discussed earlier, the reanalysis from lexical noun to lexical verb involves a process of zero-copula elimination and noun to verb reanalysis.6 Genitive use of empunya may have emerged partly via analogy between the nominal domain and the clausal domain. As highlighted across (23b) and (23c), the agent role in subject position (i.e. hamba ‗I‘) is reinterpreted as the possessor noun, while the patient role in object position (i.e. kerajaan ‗kingdom‘) is reinterpreted as the possessee noun, with lexical verb empunya (‗possess‘) reinterpreted as the genitive linker. The availability (at least in contemporary imitations of Classical Malay) of utterances such as (24) provides clear evidence that empunya has grammaticalized into a genitive or possessive marker. (24) Bukannya iai [ hamba empunya kerajaani ] NEG.3SG 3SG 1SG.HUMBLE GEN kingdom ‗It‘s not that that is my kingdom‘ (focused interpretation) ‗It‘s not that it‘s my kingdom‘ (non-focused interpretation) Note that in this example, genitive (em)punya appears as part of a possessive construction (hamba empunya kerajaan ‗my kingdom‘) that is embedded within a cleft-type construction marked by negative polarity focus marker Bukannya (‗It‘s not that …‘), as in (23) above. In addition, it also appears within an equi-construction that highlights the possessee noun. This is done with a third person singular pronoun ia (‗it‘, ‗that‘), which is co-referential with the possessee noun kerajaan ‗kingdom‘, is introduced in focus position immediately following cleft-marker Bukannya. This move then effectively blocks any interpretation of the expression hamba empunya kerajaan as a finite clause. That is, we no longer obtain the interpretation: ‗I am the possessor/sovereign of the kingdom‘. Consequently, the first person singular pronoun hamba ( ‗I‘) cannot now be interepreted as a subject NP, but rather is reanalyzed as possessor NP in a genitive construction. The emergence of genitive punya thus marks a subtle shift in focus, with emphasis on the descriptive content of lexical noun empunya (i.e. ‗possessor‘) giving way to emphasis on the logical content of genitive punya (i.e. ‗possessor-possessee relationship‘). Clear evidence of this shift is seen in (24) above, where third person singular pronoun ia (‗it‘) is co-referential with the possessee noun kerajaan (‗kingdom‘) rather than the possessor noun hamba (‗I‘).


4.3 Emergence of pronominal (em)punya A natural development intimately linked with the rise of genitive punya is the emergence of possessive pronominal punya. Essentially, pronominal uses of punya are context-dependent and emerge quite simply—that is, in the absence of overt expression of the possessee NP, a ‗headless‘ nominal construction obtains, as highlighted in (25).7 (25) Bukannya hamba empunya Ø

‗It‘s not mine.‘ 4.4 Emergence of stance punya The link between genitive punya and stance punya is somewhat less direct. Structurally, whereas genitive punya involves a nominal construction, sentence-final stance punya involves instead a clausal construction, arguably nominalized. This structural difference is highlighted in (26) and (27) below.8 (26) Ini aku punya tulisan. this 1SG GEN writing ‗This is my (hand)writing.‘ akan bayar punya. pay STANCE ‗He won‘t pay (I can assure you).‘ As seen in (26), genitive punya constructions involve the overt expression of a head noun—more specifically, the possessee NP. On the other hand, as seen in (27), sentence-final stance punya constructions are ‗headless‘. If a link exists between these two kinds of punya constructions, pronominal (em)punya constructions such as (28) appear to be crucial as a bridge construction. This is because, in sentence-final position, possessive pronominal punya can easily host the typical end-of-utterance prosodic stress accompanying the speaker‘s assertion. This explains to some extent why punya is ideal as a possessive pronominal in contrastive contexts that emphasize possessor identity, as noted earlier in §3.3. (28) Ini aku punya ø. this 1SG POSS-PN ‗This is mine.‘


Dia tak

Crucially, for the development of stance punya, pronominal punya has extended into nonpossessive contexts, as in (29) below, where the utterance Aku nak (yang) merah punya can be interpreted either as ‗I want (the) red one(s)‘ or ‗I want (the) red ones, mind you‘. Punya in such utterances resembles Singlish (Singapore English) pronominal one, which can also be used in sentence-final position as a speaker mood marker (typically assertive, as in He won‘t talk to you one, meaning something like ‗Don‘t expect him to talk to you; trust me, I know him‘). In this regard, both Malay punya and Singlish one allow for a pronominal as well as speaker mood or stance marker interpretation. As seen in (29), emphasis on speaker mood yields reanalysis of punya from pronominal to nominalizer or stance complementizer usage, particularly in the presence of definiteness marker yang, which readily induces a nominal interpretation for merah (‗red‘) even in the absence of punya as pronominal marker. (29) Aku nak (yang) merah punya ø.


want NOMZ red PN/STANCE a. ‗I want the red one(s).‘ b. ‗I want the red ones, mind you.‘

Worth noting is that pronominal one in Singaporean English can be pluralized, when definite referents are involved (cf. indefinite pronoun someone/*someones), while sentencefinal one appears only in the singular form. This morphological constraint also suggests that sentence-final one is more grammaticalized than pronominal one. Malay punya, on the other hand, does not have a plural counterpart. Thus, whereas pluaralizable English pronominal one(s) is still highly nominal and hence fairly lexical, non-pluralizable Malay punya is relatively more grammatical. Not surprisingly, the distinction between pronominal punya and sentence-final stance punya is often subtle, given the absence of morphological cues, as well as the absence of clear differences in terms of linear word order. Nevertheless, being sentence-final, stance punya can host a wide range of end-of-sentence speaker prosody, which often belies much about a speaker‘s mood. To briefly sum up our discussion thus far, in terms of motivation, at least two languageinternal factors converge to facilitate the development of punya as a stance marker. One important factor is the head-final position of pronominal punya, which readily attracts prosodic features such as emphatic stress as well as other emotive intonation contours (e.g. assertion, resignation, doubt). Another important factor is the availability and frequent use of nominalizer/complementizer yang, which can readily assume a definiteness-marking function, leaving punya to primarily serve as stance marker. In the following section, we examine a language-external factor (namely, contact with southern Chinese dialects) which has also contributed to the emergence of stance and other new grammatical functions of punya. 5. Language contact and the grammaticalization of punya 5.1 Evidence of language contact As mentioned earlier, use of punya as a genitive and possessive pronominal became highly productive in 19th century texts. This phenomenon coincided with a massive wave of Chinese immigrant workers brought in to work in the tin mining states of peninsular Malaya (now West Malaysia) around the middle of the 19th century.9 The Chinese dialects spoken by these immigrant workers also have grammatical morphemes that are similar to punya in terms of genitive and possessive pronominal functions. Among these morphemes are Cantonese ge3, Hakka ke, Hokkien e, and Chaozhou kai. Structural similarities between genitive and possessive pronominal punya and their Chinese equivalents are highlighted in (30) and (31) below, using Cantonese examples for illustrative purposes. (30) a. Malay b. Cantonese engkau punya seteru nei5 ge3 dik6jan4 2SG GEN enemy 2SG GEN enemy ‗your enemy‘ (IB 4:6, 19th c.) ‗your enemy‘ (31) a. Malay Ini aku punya. this 1SG GEN ‗This is mine.‘ (PA 111:8, 19th c.) b. Cantonese nei1go3 ngo5 ge3 this:CL 1SG GEN ‗This is mine.‘


There is strong reason to believe that the observed similarities in structure and function between Malay punya and the genitive/pronominal marker in Cantonese and other southern Chinese dialects played a vital role in the copious increase in genitive and possessive pronominal punya constructions in the mid-19th century.10 Evidence of contact-induced grammaticalization can be seen in the rise of a Bazaar-type usage (i.e. ‗market variety‘) of punya in associative and relative clause constructions, as shown in (32b) and (33b) respectively, both of which differ from their counterparts in standard Malay and Cantonese. These Bazaar-type punya constructions flourished for a period of time, attested in 19th century texts and persisting into 20th century Malay until the effects of the language standardization policy of the 1970‘s began to be felt even within the colloquial domain.11 (32) a. Standard Malay Orang India (memang) suka makan pedas. person(s) India really like eat ‗Indians (of course) like eating spicy food.‘ b. Bazaar Malay India punya orang (banyak) suka makan pedas. India ASSOC person(s) plenty like eat ‗Indian people like eating spicy food (very much).‘ Cantonese Jan3dou6 jan4 (zan1hai6) zung1ji3 sik6 laat6 je5. India person(s) really like eat thing ‗Indian people (really) like eating spicy food.‘ Standard Malay Pinggan (yang) saya guna/*makan belum di-cuci (lagi). plate REL 1SG use/eat not-yet PASS-clean yet ‗The plate that I used/ate (from) has not been washed (yet).‘ Bazaar Malay Saya guna/ makan punya pinggan belum cuci (lagi). 1SG use/eat REL plate not-yet clean yet ‗The plate I used/ate (from) has not been washed (yet).‘ Cantonese Ngo5 jung6/sik6 gwo3 ge3 wun2 zong6 mei6 sai2 gaa3. 1SG use/eat EXP REL bowl still not-yet wash SFP ‗The bowl I used/ate (from) has not been washed (yet).‘






Worth noting is that native speakers of Malay typically use postnominal modifying constructions among themselves, as illustrated in (32a) and (33a) above. However, in contact situations these same native speakers frequently use Bazaar-type prenominal modifying constructions such as (32b) and (33b), largely to accommodate to their non-native speaker interlocutors. Note that modifying constructions in the Chinese dialects are also prenominal, as illustrated in (32c) and (33c). There is some evidence from the usage of Malay punya that suggests native speakers do not blindly imitate a contact language, but instead are sensitive to the constraints of their


native language as well. Consider the use of associative punya constructions such as (32b) above, where we see evidence of an ‗interlanguage‘ construction that conforms to neither Malay (the native language) nor any of the southern Chinese dialects (the contact languages). That is, whereas the canonical associative construction in Malay is [N][XP], as in (32a), and that of the Chinese dialects is the converse [XP][N], as in (32c), Bazaar Malay compromised by creating an intermediate bridge construction using an [XP] punya [N] construction, as in (32b), thus forming a prenominal associative construction of non-native origin.12 Apparently, a Chinese-type [XP][N] construction produces an alignment that is diametrically opposite to that of the canonical [N][XP] construction in Malay, hence a blatant violation that is not tolerated unless marked in some fashion, in this case through the agency of overt associative marker punya. There are also other indicators of an interlanguage system in the Bazaar-type usage of punya constructions. As seen in the contrastive pair in (32a/b), there is evidence of nonnative lexical choices, such as the use of quantifier banyak ‗plenty‘ as an intensifier when modifying the predicate suka ‗like‘ in Bazaar Malay (as opposed to the use of memang ‗truly‘ in Standard Malay). There is also evidence of deficient verbal morphology, as in (33b) where the passive di- prefix is missing. It is also interesting to note that native speakers of Malay sometimes use a ‗sing-song‘ prosody in the company of non-native speaker interlocutors, imitating to some extent their Chinese interlocutors whose speech is influenced by their tonal native languages, and extending this speech style to interlocutors from other ethnic and linguistic backgrounds, yet this peculiar ‗sing-song‘ style is not carried over into conversations with fellow Malay native speakers. This is reminiscent of ‗foreigner talk‘ strategies. We thus see here attempts of interlanguage structuring, not only at the lexical and morphosyntactic levels, but also at the level of prosody. Also interesting is the appearance of [yang ... punya] relative clauses such as (34). As highlighted in (34‘), such constructions comprise of a nominalized predicate (e.g. yang saya guna ‗(the one) that I used‘) being linked to the head noun (e.g. pinggan ‗plate‘) by relativizer punya. Note that the presence of punya as linker is obligatory, indicating that prenominal clausal modification comes at some cognitive cost. (34) Yang saya guna *(punya) pinggan belum di-cuci (lagi). NOMZ 1SG use REL plate not-yet PASS-clean yet ‗That plate which I used has not been washed (yet).‘
(34‘) [NominalizedClause Yang saya guna] [REL punya ][HeadNoun pinggan] ... [NominalizedClause The one I used] [REL PUNYA ][HeadNoun plate] ...

Such [yang ... punya] relative clause constructions are reminiscent of Classical Tibetan (as well as Lhasa Tibetan) relative clauses discussed in DeLancey (1986, 1989, 2003), where a nominalized predicate (note the use of nominalizer suffix –pa) is linked to the head noun using genitive –i as linker, as illustrated in (35) and (36) below. (35) Classical Tibetan: sgas pa-‗i zas rnams gather NOMZ-GEN food PL ‗the food which had been collected‘ (DeLancey 2003:264) (36) Lhasa Tibetan:


mo-s bzos-pa-‗i mog=mog she-ERG cook-NOM-GEN momo ‗the momos which she made‘

(DeLancey 2003:276)

Equally interesting is that the predicate nominalized by yang is native to the Malay language, and in canonical relative clause constructions, as highlighted in (37) below, this nominalized predicate appears in post-nominal (not pre-nominal) position, consistent with the canonical [N][XP] template. Although syntactically subordinate, the structural link to an appositive origin is highly transparent for postnominal-type relative clauses.
(37) [HeadNoun pinggan] [RelativeClause yang saya guna] ... [HeadNoun plate] [RelativeClause that I used] ...

An appositive link is less transparent when a predicate nominalized by yang appears in pre-nominal position. To begin with, an appositive construction with prenominal [yang + XP] modifying a head noun is not permissible in the Malay language, as highlighted in (38) below. This prohibition is explainable in cognitive processing terms. As noted in the literature (Hawkins 1994, 2004; Kwan 2004, 2005; Matthews & Yeung 2001; Cheung 2006; see also §5.2.1), deploying head-initial modifiers in pre-nominal modifying constructions is not costefficient for SVO languages. Indeed, among SVO languages, only Chinese dialects are known to make extensive use of prenominal modification (Dryer 1992). Such prohibitive cognitive cost may partly explain why the recruitment of punya as linker is obligatory. This would also help explain why native speakers find [yang ... punya] prenominal relative clause constructions such as (39) more acceptable than punya-less constructions such as (38).
(38) (39) *[NominalizedClause Yang saya guna] [HeadNoun pinggan] ... *[NominalizedClause The one I used] [HeadNoun plate] ... [NominalizedClause Yang saya guna] [REL punya ] [HeadNoun pinggan] ... [NominalizedClause The one I used] [REL PUNYA ] [HeadNoun plate] ...

Much more common, especially in colloquial Malay, are ‗headless‘ [yang ... punya] constructions, such as (40). This headless construction arguably involves reanalysis of punya from relative clause marker to appositive pronominal, yielding a patient nominal yang saya guna punya (lit. ‗that which I use one‘, reminiscent of the appositive form ‗that which I use, that one‘), as highlighted in (40‘). (40) Yang saya guna punya belum di-cuci (lagi). NMLZ 1SG use NMLZ not-yet PASS-clean yet ‗The one that I used has not been washed (yet).‘
(40‘) [NominalizedClause Yang saya guna] [Pronominal punya [HeadNoun  ]] ... [NominalizedClause The one I used] [Pronominal one [HeadNoun  ]] ...

Also interesting is the equivalence of punya in (40) above to distance demonstrative (i)tu ‗that‘ in (41) below, which highlights a strong pronominal reading for punya. (41) Yang saya guna (i)tu belum di-cuci (lagi). NMLZ 1SG use that not-yet PASS-clean yet ‗The one that I used has not been washed (yet).‘
(41‘) [NominalizedClause Yang saya guna] [Pronominal (i)tu [HeadNoun  ]] ... [NominalizedClause The one I used] [Pronominal one [HeadNoun  ]] ...


As seen in (42) and (43) below, Cantonese does not have a yang-type nominalizer in relative clauses (whether headed or headless). That is to say, Cantonese prenominal modifying constructions do not make use of a nominalizer in clause-initial position. Instead we see the use of either relative clause linker ge3, as in (42), or nominalizer ge3, as in (43)— the latter in clause-final position.13 These constructions are similar to Malay punya relative clauses (headed and headless). (42) Ngo5 jung6 gwo3 *(ge3) wun2 (zong6) mei6 sai2. 1SG use ASP REL bowl still NEG wash ‗The bowl that I used has not been washed (yet).‘ (43) Ngo5 jung6 gwo3 *(ge3) (zong6) mei6 sai2. 1SG use ASP NOMZ still NEG wash ‗The bowl that I used has not been washed (yet).‘
[ModifyingClause ngo5 jung6 gwo3] [REL ge3] [HeadNoun wun] ... [‗Stand-alone‘Clause ngo5 jung6 gwo3] [NOMZ/STANCE ge3 [NP Ø]] ...

(42‘) (43‘)

It is possible, in Cantonese, to produce utterances such as (44), where go2go3 ‗that‘ appears either as preposed object for the matrix clause (ngo5) (zong6) mei6 sai2 ‗(I‘ve) (still) not yet washed (it)‘, or as definiteness marker for the head noun wun ‗bowl‘ in the embedded (subordinate) relative clause ngo5 jung6 gwo3 ge3 wun ‗the bowl that I used‘. However, Cantonese demonstrative go2go3 has not (yet) grammaticalized into a nominalizer, unlike Malay yang. (44) Go2go3 ngo5 jung6 gwo3 *(ge3) wun2 (zong6) mei6 sai2. that 1SG use ASP REL bowl still NEG wash ‗That bowl that I used (still) has not been washed (yet).‘ To briefly sum up this section, we have seen some important similarities between Malay punya and Cantonese ge3 constructions, particularly in structures with nominal modification functions. At the same time, we have also seen differences which suggest that grammatical constructions that emerge as a result of language contact often accommodate the constraints of both native and contact languages. 5.2 Further evidence of language-specific constraints While contact-induced grammaticalization underscores strong parallels between the native language and the contact language(s), language-specific constraints at the same time can often give rise to variance. One such variance was noted above with respect to the appositivetype [yang ... punya] constructions in Malay, for which we find no equivalence in Cantonese ge3 constructions. In what follows we discuss additional examples that show variance. 5.2.1 Emergence of pre-adjectival stance constructions As discussed earlier in §3.4 and §4.3, Malay punya has also evolved pre-adjectival stance functions within the colloquial register, as in (45). Worth noting is that native speakers interpret this as an extension of begitu punya constructions.14 As seen in (46), begitu punya conveys the meaning ‗to that incredible extent‘, and the utterance reflects a strong sense of speaker subjectivity. In general, with (begitu) punya + ADJ constructions, speaker mood can


take various shades of incredulity, ranging from amazement and awe to sarcasm and contempt. (45) Orang India punya pandai makan pedas. person(s) India INTSF clever eat ‗Indians really are good at eating spicy food.‘ (46) Orang India begitu punya pandai makan pedas. person(s) India like.that LNK/INTSF clever eat Lit. ‗Indians are good at eating spicy food to that extent.‘ Intended meaning: ‗Indians really are good at eating spicy food.‘

This type of intensifier construction is generally not available in Cantonese and other southern Chinese dialects. That is, as seen in (47) below, Cantonese has not extended the use of ge3 to pre-adjectival stance uses. Rather, ge3 more typically yields an assertive stance interpretation primarily in sentence-final position, sometimes accompanied with other sentence-final particles such as wo5, as in (48).15 As seen in (49) below, and discussed earlier in §3.4 and §4.3, Malay has recourse to a similar strategy involving sentence-final stance punya/mya. (47) Cantonese Jan3dou6 jan4 (zan1hai6/*ge3) lek1 sik6 laat6 (wo5). India person(s) really clever eat SFP ‗Indians are really good at eating spicy food.‘ (48) Cantonese Jan3dou6 jan4 (zan1hai6/*ge3) lek1 sik6 laat6 ge3 (wo5). India person(s) really clever eat STANCE SFP ‗Indians are really good at eating spicy food.‘ Colloquial Malay Orang India (memang) pandai makan pedas punya/mya. person(s) India really clever eat STANCE ‗Indians (of course) like eating spicy food. (No doubt about it.)‘


The mood-modulating effects of Cantonese ge3 and Malay punya/mya in sentence-final position, however, are qualitatively different from those of pre-adjectival (begitu) punya constructions in Malay or intensifier-type zan1hai6 ‗really‘ constructions in Cantonese. While pre-adjectival stance punya serves an intensifying function at the predicate level, utterance-final punya/mya encodes speaker stance at the propositional level, and its effect can be either intensifying (yielding a sense of assertion) or, depending on the accompanying prosody, mitigating, conciliatory, softening, hedging, even doubting and challenging. It is worth noting that this difference in pragmatic effect correlates with subtle differences in type of reanalysis—with (begitu) punya + ADJ environments facilitating a linker-to-intensifier reanalysis, as in (45) and (46), and with utterance-final environments facilitating a nominalizer-to-stance particle reanalysis, as in (49). While Cantonese ge3 does not appear in pre-adjectival stance position the way Malay punya does, there is evidence of ge3 in pre-nominal stance constructions such as nei1 di1 gam2 ge3 je5 ‗this sort of thing‘, or go3 gam2 ge2 je5 ‗that such thing‘. The latter usage


(note the tonal modification from ge3 to ge2, i.e. a change from a level tone to a rising one) is much more restricted and conveys a strong sense of disapproval or disgust (Chan, Tsui & Shin 2003). The examples in (50) below highlight the shift in stance as ge3 is reduced to ge2. (50) a. Keoi5 m4 soeng2 gin3 go2 di1 gam2 ge3 je3. 3SG NEG want see that CL.PL such LNK thing ‗She does not want to see that sort of thing (< those types of things).‘ b. Keoi5 jau6 soeng2 gin3 go2 (gam2) ge2 je5? 3SG again want see/meet that such STANCE thing ‗(You mean), she wants to see that scum yet again?‘

As discussed in Chan et al. (2003), the stance reading comes from intensifier gam2. This argument is tenable on both semantic and phonological grounds, since the meaning that ge2 now conveys is precisely that of gam2, and since we clearly also see evidence of tone spreading as ge3 changes to ge2 under the influence of gam2. The derogatory reading of ge2 is unmistakable in utterances such as (50b), where both prosodic lengthening and tonal reduction can further yield a ge32 stance particle that produces a sarcastic effect, in tandem with the deployment of an inanimate noun je5 ‗thing‘ to refer to someone that the speaker is very upset with and/or despises (to the extent that this other person is being referred to as ‗that (slimy) thing‘, ‗that low heel‘ or ‗that scum!‘). Even in Cantonese, the range of nouns that can be modified by derogatory ge2 is highly restricted, involving mainly inanimate nouns such as je5 ‗thing‘ to refer to humans. As seen in (51c) below, Malay punya does not have this derogatory-type prenominal modifying function. (51) a. Aku tak mau dengar benda [macam itu]. 1SG NEG want hear thing like that ‗I don‘t want to hear things of that sort / such things.‘ b. ? Saya tak

mau dengar [itu macam punya] benda. want hear like.that LNK thing ‗I don‘t want to hear things of that sort / such things.‘



mau dengar [punya] benda. want hear STANCE thing ‗I don‘t want to hear such things.‘


Several factors impede the emergence of stance punya constructions in pre-nominal position. To begin with, unlike the Chinese dialects, the Malay language has a strong preference for post-nominal modification, as noted earlier in §2, and illustrated in (51a) above. A related reason, as highlighted in (51b), is the marginalization of pre-nominal modifying constructions to colloquial usage and to non-standard varieties such as Bazaar Malay. Both these factors combined account for the somewhat tenuous recognition given to punya in pre-nominal relative clause contexts. Crucially, without developing a robust relative clause linker function, punya is not strategically positioned to evolve a pre-nominal stance function. As noted earlier, this is related to the prohibitive cognitive processing cost for SVO languages to recruit an adnominal-type grammatical construction, particularly relative clause, since these often involve lengthier (hence ‗heavier‘) constituents in specifier position.


To elaborate further, recall that among SVO languages, only Chinese dialects have thus far been identified with prenominal modification (Greenberg 1966; Dryer 1992). In addition, as predicted by cognitive processing models (e.g. Hawkins 1990; 1994; 2004), evidence from Cantonese reaction time studies indicate that prenominal relative clauses are processed slower than topicalized clauses, with performance declining as the relative clauses get longer (Matthews & Yeung 2001; Cheung 2006).16 Recall that the use of ge3/ge2 in prenominal position to express feelings of surprise, reservation, or disapproval is rather restricted even in Cantonese, and this rare phenomenon has not been reported for other Chinese dialects.17 It is thus not surprising that the grammaticalization of Malay punya has not extended into this direction. To briefly sum up this section, we have looked at some examples of language-specific variations in the grammaticalization of Malay punya and its equivalent in southern Chinese contact languages as illustrated through Cantonese ge3. In particular, we have focused on differences in the way pre-nominal modifying constructions are occasionally realized across these languages. In addition, we have examined differences in the types of attitudinal specification that Malay punya and Cantonese ge3 can handle and have found an asymmetry in the availability of pre-adjectival and pre-nominal stance constructions. Overall, we have seen how language-specific properties (including structural ones) influence the range of grammaticalizable categories for each language in contact situations. In the next section we will focus on another type of modification, namely subordination, which operates at the level of clausal modification. This is distinct from the genitive, associative and relative clause uses of punya which operate at the level of nominal modification, and unlike sentence-final stance punya which operates at a more socio-interactive level involving a syntactic-pragmatic interface to convey a speaker‘s (inter)subjective stance toward a proposition. 5.2.2 Evidence of etymological constraints on the emergence of subordinator functions Evidence from contemporary Malay suggests that the emergence of subordinator punya is closely linked to the availability and semantic extension of genitive punya. Although now marginalized in Standard Malay, pre-nominal genitive punya constructions still continue to flourish in casual conversation and informal writing. A random quick-search on the internet readily yields examples such as (52) and (53) below.18 (52) Brapa dia punya harga? how.much 3SG GEN price ‗How much is its price?‘ (53) ia

masuk syurga enter heaven



punya pasal matter/business/affair/problem/etc.


masuk neraka ia punya pasal enter hell 3SG GEN matter/business/affair/problem/etc.

‗(If) he goes to heaven, (it‘s) his business; (if) he goes to hell, (it‘s) his business (as well).‘ Use of punya in conjunction with pasal (a general or ‗formal‘ noun used in a wide range of contexts to express situational meanings such as ‗matter‘, ‗business‘, ‗affair‘, ‗problem‘, ‗fault‘, ‗cause‘, ‗reason‘, etc.) was found to be especially productive in the informal register


that characterized open forum discussions over the internet. In terms of semantic extension, we can easily see evidence of genitive punya in dismissive (and sometimes blame-assigning) utterances such as (53) above extending into causal connective contexts such as (54) to (55) below. Pragmatic inferencing plays an important facilitative role (Traugott 1988). That is, when a reader/hearer encounters juxtaposed clauses, he or she will naturally try to infer possible semantic relations between the two clauses. In the case of preposed punya pasal constructions, we obtain subordinate clauses of cause/reason. (54) Cemburu punya pasal, kekasih jadi mangsa. Jealous LNK reason beloved become victim ‗Because (he was) jealous, his beloved died.‘ (55) Dok dia

panik punya pasal PROG panic LNK reason terus emergency brek straightway emergency brake

‗Because (s/he was) panicking, s/he immediately applied the emergency brake.‘ 19 As seen in (54) and (55), the use of punya with adjectival and clausal specifiers gives rise to the reanalysis of punya pasal as a causal connective—more specifically, as a clause-final causal subordinator (equivalent in meaning to clause-initial because in English). One way to chart this reanalysis is to consider the phenomenon in terms of a weakening in syntactic restrictions, as illustrated in (56) to (58) below.20 (56) Dia punya pasal, kekasih jadi mangsa. 3SG GEN fault beloved become victim Lit. ‗His fault, his beloved died.‘ ‗Because of him, his beloved died.‘ (57) Cemburu punya pasal, kekasih jadi mangsa. jealous(y) LNK reason beloved become victim ‗Because of jealousy/Because he was jealous, his beloved died.‘ Dok panik punya pasal, kekasih jadi mangsa. PROG panic LNK reason beloved become victim ‗Because he was panicking, his beloved died.‘
[NP dia][GEN punya][NP pasal], [ Matrix clause] [ADJ cemburu][Causal subordinator punya pasal], [ Matrix clause] [VP dok panik][Causal subordinator punya pasal], [ Matrix clause]


(56‘) (57‘) (58‘)

Structurally, we see the head noun pasal ‗cause/reason‘ combining with the linking morpheme punya to form a new grammatical marker punya pasal, which functions as a genitive-based cause/reason marker in (56). In this usage, it is equivalent to a ‗because of‘ preposition-type cause/reason marker in English. This form is then is further reanalyzed as a cause/reason subordinator in (57) and (58). In both cases, these punya pasal constructions


identify the reason or cause for the situation that is being described in the upcoming matrix clause. A process of structural simplification is involved here, in which a genitive construction of the type [ModifyingPhrase XP][Linker punya] [NP pasal]—such as dia punya pasal ‗his fault‘ in (56‘)—is reanalyzed into a routinized, structurally simpler and cognitively more processible construction [ModifyingPhrase XP][Subordinator punya pasal]—as in (57‘) and (58‘), with a more direct reading equivalent to English subordinator because of as in ‗because he was jealous / because he was panicking‘, instead of the prepositional phrase usage ‗because of jealousy / because of his panicking‘. Thus, consistent with previous claims in the literature which have identified possessors and subordinate constructions (e.g. conditionals) as topic constructions (e.g. Haiman (1978) and Taylor (1996); see also discussions on ‗mental space builders‘ in Fauconnier (1985), Fujii (1995), Sweetser and Fauconnier (1996), and Dancygier and Sweetser (2005)), punya pasal constructions serve an important cognitive function in helping the addressee to more easily identify an intended referent (in this case, the affected outcome) that forms the focal point of discussion in the subsequent main clause. Further worth recalling is that punya pasal cause/reason constructions often convey strong subjective evaluations on the part of the speaker, including the assignment of blame. This stance reading emerges quite naturally, given that possessors and clausal subjects tend to be strongly associated with agentive or causative properties, and therefore can generally be held responsible for consequences of their actions, or even inactions, these consequences being more explicitly spelled out in the matrix clause—as in ‗because of him/because he was panicking, someone died.‘ Cantonese ge3 has also evolved topic marking as well as clause-subordinating functions, as illustrated in (59) and (60) respectively. Both functions involve a ge3waa2 construction, in which a relative-type [XP][LNK ge3 [NP waa2]] construction is reanalyzed into a subordinatortype [XP][Subordinator ge3waa2] construction. A notable result is structural simplification. Etymologically, waa2 appears to have been derived from a lexical noun waa6 meaning ‗word(s) or speech‘ (which has a verbal counterpart waa6 meaning ‗say‘).21 Hence its close association with (‗Speaking of XP, ...‘) topic constructions and (‗Say (i.e. If) XP, ...‘) conditional clause constructions. (59) (gong2dou3) ngo5 (ge3) lou3baan2 ge3waa2, speaking.of 1SG GEN employer TOP keoi5 nei3 fan6 jan4 hou2 gu3zap1 ge3. 3SG this type person very stubborn STANCE ‗Speaking of my employer, he‘s a very stubborn kind of person.‘ ‗As for my employer, he‘s a very stubborn kind of person.‘ (60) (jyu4gwo2) hai6 ngo5 ge3waa2, say/if COP 1SG COMP ngo5 zau6 1SG then m4

wui5 gam2 zou6 lak6. FUT like.this/that do SFP

‗Say it were me, I wouldn‘t do it (this/that way).‘ ‗If it were me, I wouldn‘t do it (this/that way).‘


An interesting crosslinguistic difference is that Malay punya pasal has evolved into a cause/reason subordinator, while Cantonese ge3waa2 has evolved instead into a topic marker, as in (59), and a conditional subordinator, as in (60).22 This difference in their grammaticalization trajectory can be traced to the etymology of their ertswhile head nouns, with Malay pasal derived from a noun meaning ‗matter/reason/cause‘ and Cantonese waa2 from a verb/noun meaning ‗say/saying (as in words or speech)‘. As we will see in the next section, crosslinguistic evidence suggests that subordinator functions often emerge via the agency of adnominal (or linking) morphemes (e.g. genitives and relative clause markers) that have further grammaticalized into nominalizers (i.e. complementizers). 6. Crosslinguistic similarities In the previous section we have examined the influence of language contact on the grammaticalization of Malay punya, and in particular we have identified similarities and differences in the semantic extensions between Malay punya and Cantonese ge3. Their grammaticalization pathways are summarized in Figures 2 and 3 below (see Sio (2007) and Yap & Matthews (in press) for further discussions of Cantonese ge3). While these pathways show language-specific variations in terms of some of their extension paths, including type of subordinator functions (as discussed in §5.2.2 above), it is clear that both Malay empunya and Cantonese ge3 share much in common in terms of their pronominal > nominalizer > stance development. In the rest of this section, we will show that this pathway is a robust one across neighboring Asian languages.
Figure 2. Grammaticalization pathways of Malay empunya Lexical noun empunya Lexical verb empunya Cause/Reason subordinator punya pasal Genitive (em)punya Possessive pronominal punya Relativize clause marker punya (colloquial register & Bazaar Malay— influenced by language contact) Nominalizer punya (headless relative clause construction— often with head-initial nominalizer yang) Stance punya

Figure 3. Grammaticalization pathways of Cantonese classifier ge3 Topic marker (gong2dou3) NP ge3waa2 Conditional subordinator (jyu4guo2) ... ge3waa2 Classifier go3 Genitive go3/ge3 Relativize clause marker ge3


Possessive pronominal ge3 Nominalizer ge3 (inclusive of headless relative clause usage) Cleft hai6...ge3 (+ SFP) Stance ge3 (+ SFP)

Other southern Chinese dialects also have morphemes that exhibit overlapping functions with Cantonese ge3 (and Malay punya). For most of these dialects diachronic accounts are rather sketchy, but emerging cross-dialectal comparisons are providing some interesting insights into the grammaticalization pathways of their nominalizers, inclusive of languagespecific variations. Matthews and Xu (2002), for example, have identified that among the southern Min dialects, the Chaozhou classifier kai not only displays virtually all the functions of Cantonese ge3, including the adnominal functions (e.g. genitive, associative, and relative clause marking), as well as cleft and sentence-final stance functions, it also appears to have evolved a non-verbal copula function as well (see also Stassen 1997; Yap & Matthews, in press). Here we illustrate with examples from another southern Min dialect, namely Hokkien. Hokkien has a classifier e, shown in (61a), which can combine with a demonstrative in the absence of the head noun (‗headless‘ contexts) to yield a pronominal, as in (61b). (61) a. tsi e gina this CL child ‗this child‘ b. tsi e this CL ‗this (one)‘

This morpheme e is also used as a genitive, as in (62a), and a relative clause marker, as in (63a). In ‗headless‘ contexts, i.e. when the lexical head noun that is being modified is omitted, the linking morpheme e is further reanalyzed as a pronominal—more specifically, as possessive pronominal, as in (62b), or as sentential pronominal or nominalizer, as in (63b). (62) a. gua e tse 1SG GEN book ‗my book‘ b. gua


‗mine‘ (63) a. gua aboe khua i hoo li e tsiupio 1SG not.yet see 3SG give 2SG REL watch ‗I have not yet seen the watch that he gave you.‘ gua


aboe khua not.yet see


hoo li e give 2SG NOMZ


‗I have not yet seen the one that he gave you.‘ The morpheme e also frequently appears as a sentence-final particle (SFP) in cleft and stance constructions, as seen in (64) and (65) respectively. (64) gua si ku ni lai e 1SG FOC(<COP) last year come SFP ‗It was last year that I came.‘ (65) gua

ku ni lai e last year come SFP ‗(It‘s that) I came last year.‘

As noted in Shinzato (2005) and elsewhere, cleft constructions inherently reflect speaker mood—particularly the speaker‘s viewpoint in terms of which argument is to be given greater discourse prominence. In this sense, cleft constructions can be said to be inherently stance-ful. Stance constructions such as (65), however, are much more subtle in conveying various shades of speaker moods, largely as a result of elision of the focus particle si, and in its place providing a greater role for nominalizer e, which now is reanalyzed as a sentencefinal particle with a default assertive interpretation, but which also allows for other mood interpretations when accompanied by marked prosody and/or other sentence-final particles. Figure 4 below summarizes the grammaticalization pathways for Hokkien classifier e, in particular highlighting its pronominal > nominalizer > stance development.
Figure 4. Grammaticalization pathways of Hokkien classifier e Demonstrative pronominal tsi e ‗this (one)‘ Classifier e Genitive e Possessive pronominal e Relativize clause marker e

Nominalizer e (inclusive of headless relative clause usage) Cleft si... e Stance e (+SFP)

In general, the pronominal > nominalizer > stance development appears to be very robust among the Chinese dialects; this phenomenon may even be pan-Chinese. Evidence from classical and contemporary Chinese (e.g. Wang 1980; Lu 1943; L. Jiang 1999; S. Jiang 2005; Shi & Li 2002) suggests a very similar grammaticalization pathway for nominalizer di/de despite differences in lexical origin (see Figure 5; see also Yap & Matthews, in press).
Figure 5. Grammaticalization pathways of Chinese nominalizer di/de Influence of zhi and zhe Genitive di/de (replacing genitive zhi)


Topic marker NP de hua (some Mandarin varieties) Conditional subordinator (ruguo) VP/IP de hua, … Lexical noun di (‗bottom‘) Pronominal di Possessive pronominal di Relativize clause marker di/de

Nominalizer di/de (inclusive of headless relative clause usage) Cleft shi ... de (+SFP) Stance de (+SFP) Predicate Nominalizer de
(NP de VP)

A pronominal > nominalizer > stance pathway has also been observed in other East Asian languages, most notably in Japanese (Shinzato 2005; Yap, Matthews & Horie 2004; Yap & Matthews, in press). Diachronic analyses (see Horie 1998 for a summary) reveal productive use of genitive and pronominal no in 8th century texts (Old Japanese), as in (66) and (67) respectively, followed by emergence of nominalizer no constructions around the 16th century (Middle Japanese), among them headless relative clause and stance constructions, as shown in (68) and (69) respectively. It is generally accepted that the trigger for this sudden rise in nominalizer constructions was the collapse of the rentaikei adnominal system, which seriously undermined the kakari musubi (cleft focus) system that had been widely used in Old and Middle Japanese (Nishi 2006; Shinzato 2005; Wrona 2005). In replacing the kakari musubi focus constructions, nominalizer no constructions assumed their stance functions as well (see (70) for a contemporary example). (66) otome no toko no be ni young.girl GEN bed ASSOC side LOC ‗on the side OF the bed OF a young girl‘ (Koziki [Anthology of Old Tales],
749 AD; cited in Horie 1998)


yakusi wa tune no mo are do ... pharmacist TOP regularity PN also exist but ‗though there exists a pharmacist of the regular kind‘
(lit. ‗As for a pharmacist, there exists (a pharmacist) of the regular kind but ...‘)
(Bussokusekika, c. 753 AD; cited in Horie 1998)


suite yomu no wa like read PN TOP ‗the thing I like to read is …‘

soregasi ga

(Kyoogenki, 1660 AD; cited in Horie 1998)


… mairi tai no come:HUMBLE want STANCE


‗… (I) would like to humbly visit‘ (70)

(Toraakirabon Kyoogen,17th century; cited in Horie 1998)

Sonna fukanoo na koto iwanai no. that impossible ASSOC thing say.not STANCE ‗Don‘t say such an impossible thing.‘ (Cook 1990:432; gloss added)

The pronominal > nominalizer > stance development of no is highlighted in Figure 6.
Figure 6. Grammaticalization pathways of Japanese nominalizer no Topic marker no dewa Nominative no (in some dialects) Genitive no Pronominal no Nominalizer no (inclusive of headless relative clause uasge) Complementizer no Cleft no desu Stance no (desu) Cause/Reason subordinator no ni/de

A similar pronominal > nominalizer > stance development is also attested in Korean (Park 1999; Rhee, in press; Shin 2005; Yap & Matthews, in press). In fact, evidence from Old, Middle and Modern Korean indicates that this development repeats itself in cyclic waves, with some variation to the range and type of nominalization constructions available for each nominalizer and with occasional overlaps in functions across nominalizers (see Rhee, in press, and Yap & Matthews, in press, for recent discussions on the grammaticalization of Korean nominalizers –um, –ci and –n kes with special emphasis on their evolution into stance markers). There are also some intimations of cyclicity, or renewal, in Classical Japanese and Okinawan (e.g. Shinzato 2005; Wrona 2005), as well as Classical Chinese (e.g. Jiang 1999). Evidence of robust cyclicity in the pronominal > nominalizer > stance development in these East Asian languages suggests that inherent language-internal forces are at work. These inherent forces set in motion the process we term ‗native‘ grammaticalization. To sum up our discussion in this section, we have seen that the pronominal > nominalizer > stance development noted for Malay empunya finds robust correspondence with similar developments in other neighboring Asian languages. Indeed, investigations into nominalization phenomena in Tibeto-Burman languages point in the same direction (e.g. Matisoff 1972; Delancey 1986, 1999; Herring 1991; Noonan 1997, in press; Bickel 1999; Coupe 2006; LaPolla 2006; Morey 2006; van Breugel 2006; Simpson, in press; Watters, in press). In fact, short of the sentence-final stance development, the development from lexical noun/classifier/demonstrative to nominalizer is well attested across the languages of the world (e.g. Diessel 1999; Heine & Kuteva 2002:106-107; Hopper & Traugott 2003). What appears to be crucial for pronominal morphemes to further grammaticalize into sentence-


final stance markers is the availability of nominalizers (i.e. complementizers) in clause-final position. In other words, structural considerations clearly have an important role. In the case of an SVO language such as Malay, which already has a versatile and productive head-initial nominalizer (namely, yang), the emergence of a new head-final nominalizer (namely, punya) required some facilitation from contact languages. 7. Conclusion In this paper we analyzed the grammaticalization of Malay empunya from three closely related perspectives. Firstly, we diachronically examined evidence of ‗native‘ (or inherent) grammaticalization, whereby lexical noun empunya developed into a genitive and possessive pronominal (see §4.2 and §4.3). Secondly, we considered the facilitative influence of contact grammaticalization from southern Chinese dialects, particularly in relation to the strengthening of genitive and pronominal uses of punya, which in due course facilitated the development of stance-marking uses of punya/mya (see §5.1). Thirdly, we identified similar and recurrent pronominal > nominalizer > stance development in many other neighboring Asian languages, with fairly endemic areal similarities testifying to the robustness of semantic and pragmatic extensions along the pronominal > nominalizer > stance pathway (see §6). In this paper, we also focused extensively on the role of language contact in facilitating grammaticalization phenomena. Attention was drawn not only to ‗calqued‘ (or borrowed) constructions, but also to intermediate forms that replicate neither native nor target language constructions. Interestingly, both ‗calqued‘ and ‗interlanguage‘ forms have correlates in other languages of the world, suggesting that some stronger universal forces are at work, with different languages realizing a subset of the rich variations available for grammaticalization. For example, the yang … punya (‗double nominalizer‘) construction emerged to support an otherwise unwieldy prenominal modification structure (i.e. a non-canonical relative clause construction used in colloquial or Bazaar Malay). The use of double (or even multiple) complementizers has been observed in Singapore Creole English (SCE), as seen in the following example from Gil (2003:495): Ah Chew buy that yesterday Lisa choose that Jamil like that one. A similar phenomenon has also been observed in intermediate stages of bilingual acquisition, as seen in the following example of a Chinese-English bilingual child (Sophie 5;04;24): Daddy, I haven‘t got that Mickey Mouse, that Chloe gave me that one (Yip & Matthews 2007:30). Double nominalization has also been observed in Toqabaqita, a Cristobal-Malaitan (Oceanic) language (Lichtenberk 2007), and it is interesting to note that complementation in the form of clausal nominalization tends to play a marginal role in this language, hence perhaps the need to reinforce via double marking. Whether a typology of double nominalization will have something to say about intermediate stages of grammaticalization (with or without the influence of language contact) deserves further investigation. In addition to highlighting the interaction between native and contact-induced forces in grammaticalization, we have also examined the intimate link between nominalization and clausal subordination. In the case of Malay punya, there is evidence of extension from genitive usage (often with possessor focus) to subordinator usage (interpretable in cognitive processing terms as a discourse staging or mental space building function); in formal terms, a D-to-C development. A similar development has also been observed in Cantonese, although the precise type of subordination relationship that emerges (e.g. cause/reason, conditional,


etc.) is dependent on the source morpheme. In Japanese and Korean, there are diachronic indications of a genitive > nominalizer > subordinator development (Horie 1998; Rhee, in press; Yap & Matthews, in press). Crosslinguistic evidence thus far indicates that subordinators could emerge via semantic extensions along the genitivization-relativizationnominalization route; however, the specific mechanisms involved appear to vary, not only from language to language, but also in relation to the semantic constraints of individual source morphemes. Clearly, further investigation into many more languages is necessary to identify both robust tendencies and language-specific variations in the development of nominalization-related subordinators. There are still a number of issues related to nominalization phenomena (see Noonan, in press; Morey 2006; Yap & Matthews, in press, for recent discussions) that have not been addressed in this paper. One issue involves the relationship between nominalization and tense-aspect marking, which is not very transparent in Malay and other languages that do not inflect for tense. Another issue involves the close relationship between nominalization and copula usage, which is much less productive in Malay than in neighboring languages such as Chinese, Japanese, Korean and many Tibeto-Burman languages. Another issue involves the relationship between nominalization and converbal constructions, and its relationship (if any) with verb serialization. Future work will need to address these issues, in large part through detailed analyses of specific languages as was attempted here, but also through crosslinguistic comparison of languages with and without overt morphological realizations of these tense-aspect, copula, and converbal markings.
Acknowlegment This research is funded by Direct Grants (#2010250 and #4450128) from the Chinese University of Hong Kong. I am also deeply indebted to Ian Proudfoot and David Gil for generously sharing their data on Malay, and to Tsai Ya-ching for help with the Hokkien data. I also wish to thank the following for fruitful discussions and invaluable comments: Yang Gu, Shoichi Iwasaki, Hongyong Liu, Stephen Matthews, Seongha Rhee, Rumiko Shinzato, Andrew Simpson, Joanna Sio, Janick Wrona, and Debra Ziegeler.

ACC CL COMP DEF FOC GEN LOC NEG NOM NOMZ Accusative Classifier Complementizer Definite Focus Genitive Locative Negative Nominative Nominalizer PL PN PREF PRES PROG REL SFP SG SUF TOP Plural Pronoun/Pronominal Prefix Present Progressive Relative clause marker Sentence final particle Singular Suffix Topic marker

Classical Malay Texts
BB IA IP MM PAK SR Hikayat Bayan Budiman (Tales of Bayan Budiman) Hikayat Ibrahim ibn Adham (Tales of Abraham, Son of Adam) Hikayat Inderaputera (Tales of Inderaputera) Misa Melayu (Malay Chronicles) Pelayaran Abdullah ke Kelantan (The Voyage of Abdullah to Kelantan) Hikayat Seri Rama (Tales of Seri Rama)


References Aarons, Debra. 1994. Aspects in the Syntax of American Sign Language. PhD Dissertation, Boston University. Adelaar, K.A. 1985. Proto-Malayic: The Reconstruction of Its Phonology and Parts of Its Lexicon and Morphology. PhD Dissertation, Leiden University. Ansaldo, Umberto, & Stephen Matthews. 1999. The Minnan Substrate and Creolization in Baba Malay. Journal of Chinese Linguistics, 27, 38-68. Bickel, Balthasar. 1999. Nominalization and Focus Constructions in some Kiranti Languages. In Yogendra P. Yadava & Warren G. Glover (eds.), Topics in Nepalese Linguistics, pp. 271-296, Kathmandu: Royal Nepal Academy. Chan, Shuen-ti Roy, Wai-ming Tsui & Mikyong Shin. 2003. Cantonese ge2 in go2ge2je5 'That Such Thing': Grammaticalization and the Structure of NP. Paper presented at the 15th North American Conference on Chinese Linguistics (NACCL-15), Michigan State University, July 11-13. Cheung, Ki-Shun Antonio. 2006. Processing Factors in Language Comprehension and Production: The Case of Cantonese Dative Constructions. MPhil Thesis, University of Hong Kong. Chor, Oi-wan Winnie. 2004. A Semantic and Pragmatic Analysis of Verbal Particles in Cantonese. MPhil Thesis, University of Hong Kong. Coerts, Jane. 1992. Nonmanual Grammatical Markers. An Analysis of Interrogatives, Negations and Topicalisations in Sign Language of the Netherlands. PhD Dissertation, University of Amsterdam, The Netherlands. Cook, Haruko Minegishi. 1990. An Indexical Account of the Japanese Sentence-Final Particle no. Discourse Processes, 13, 401-439. Coupe, Alec. 2005. Teasing Apart Nominalization and Relativisation in Mongsen Ao. Paper presented at the Australian Linguistics Society Annual Conference, September 28-30. Dancygier, Barbara & Eve Sweetser. 2005. Mental Spaces in Grammar: Conditional Constructions. Cambridge, UK: Cambridge University Press. DeLancey, Scott. 1986. Relativization as Nominalization in Tibetan and Newari. Paper presented at the Nineteenth International Conference on Sino-Tibetan Languages and Linguistics, Ohio State University, Columbus, Ohio, USA, September 11. DeLancey, Scott. 1999. Relativization in Tibetan. In Yogendra P. Yadava and Warren G. Glover (eds.), Topics in Nepalese Linguistics, pp. 231-49, Kathmandu: Royal Nepal Academy. DeLancey, Scott. 2003. Classical Tibetan. In Graham Thurgood & Randy J. LaPolla (eds.), The Sino-Tibetan Languages, pp.255-269. London: Routledge. DeLancey, Scott. 2003. Lhasa Tibetan. In Graham Thurgood & Randy J. LaPolla (eds.), The Sino-Tibetan Languages, pp.270-288. London: Routledge. Diessel, Holger. 1997. The Diachronic Reanalysis of Demonstratives in Crosslinguistic Perspective. Chicago Linguistic Society 33: 83-97. Diessel, Holger. 1999. Demonstratives: Form, Function, and Grammaticalization. Amsterdam: John Benjamins. Dryer, Matthew. 1992. The Greenbergian Word Order Correlations. Language, 68, 81-138.


Englebretson, Robert. 2003. Searching for Structure: The Problem of Complementation in Colloquial Indonesian Conversation. Amsterdam: John Benjamins. Fauconnier, Gilles. 1985. Mental spaces: Aspects of Meaning Construction in Natural Language. Cambridge, MA: MIT Press. Fujii, Seiko Yamaguchi. 1995. Mental-Space Builders: Observations from Japanese and English Conditionals. In Masayoshi Shibatani & Sandra Thompson (eds.), Essays in Semantics and Pragmatics: In Honor of Charles J. Fillmore. Amsterdam: John Benjamins. Fung, Roxanna. 2000. Final Particles in Standard Cantonese: Semantic Extension and Pragmatic Inference. PhD Dissertation, Ohio State University. Gil, David. 1999. The Grammaticalization of punya in Malay/Indonesian dialects. Paper presented at the Ninth Annual Meeting of the South-East Asian Linguistics Society, University of California, Berkeley, USA, May 22. Greenberg, Joseph H. 1966. Some universals of grammar with particular reference to the order of meaningful elements. Universals of Grammar, ed. Joseph H. Greenberg 2nd edition, 73-113. Cambridge, Mass: MIT Press. Haiman, John. 1978. Conditionals are Topics. Language, 54, 564-589. Hawkins, John A. 1990. A Parsing Theory of Word Order Universals. Linguistic Inquiry, 21.2, 223-262. Hawkins, John A. 1994. A Performance Theory of Order and Constituency. Cambridge, UK: Cambridge University Press. Hawkins, John A. 2004. Efficiency and Complexity in Grammar. Oxford: Oxford University Press. Heine, Bernd & Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Herring, Susan C. 1991. Nominalization, Relativization, and Attribution in Lotha, Angami, and Burmese. Linguistics of the Tibeto-Burman Area, 14.1, 55-72. Holm, John. 1989. Pidgins and Creoles; Volume 2: Reference Survey. Cambridge, UK: Cambridge University Press. Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd ed. Cambridge, UK: Cambridge University Press. Horie, Kaoru. 1998. On the Polyfunctionality of the Japanese Particle no: From the Perspectives of Ontology and Grammaticalization. In Toshio Ohori (ed.), Studies in Japanese Grammaticalization: Cognitive and Discourse Perspectives. Tokyo: Kurosio Publishers. Jiang, Lansheng. 1999. The Genitive Use of Locative Words and the Origin of the Structural Particle di3.. Zhongguo Yuwen, 2, 83-93. Jiang, Shaoyu. 2005. Outline on Studies of Modern Chinese. Beijing: Peking University Press. Kwan, Wing Man. 2004. Placement of locative prepositional phrases in Cantonese: a processing perspective. Proceedings of the 16th North American Conference on Chinese Linguistics, pp. 170-186. Kwan, Wing Man. 2005. On the Word Order of Locative Prepositional Phrases in Cantonese: Processing, Iconicity and Grammar. MPhil Thesis, University of Hong Kong. LaPolla, Randy J. 2006. Nominalization in Rawang. Paper presented at the 39th International Conference on Sino-Tibetan Languages and Linguistics, University of Washington, Seattle, September 15-17.


Law, Sam Po. 1990. The Syntax and Phonology of Cantonese Sentence-Final Particles. PhD Dissertation, Boston University. Lee, Hun-tak Thomas. 2006. Ongoing Change in the Nominal Structure of Hong Kong Cantonese: The Perspective from Child Language. Paper presented at the Fifth Workshop on Cantonese (WOC-5): Sociolinguistic Perspectives on Recent Changes in Cantonese, Chinese University of Hong Kong, April 28. Lee, Yin-fai. 2005. Negation in Hong Kong Sign Language. MPhil Thesis, Chinese University of Hong Kong. Lichtenberk, Frank. 2007. Nominalization in Toqabaqita and Other Cristobal-Malaitan Languages. Keynote paper presented at the International Workshop on Nominalizers and Copulas in East Asian and Neighboring Languages, Chinese University of Hong Kong, January 8-12. Liddell, Scott K. 1980. American Sign Language Syntax. Mouton. Lord, Carol. 1993. Historical Change in Serial Verb Constructions. Amsterdam: John Benjamins. Lu, Shuxiang. 1943. On the Difference Between di3 and di4 and Their Origins. Jinling Qilu Huaxi Daxue Zhongguo Wenhua Congkan 3. Matisoff, James A. 1972. Lahu Nominalization, Relativization, and Genitivization. In John Kimball (ed.), Syntax and Semantics, volume 1, pp. 237-58. New York: Seminar Press. Matthews, S. & Hui-ling. Xu. 2002. Nominal Copula in Chaozhou. Paper presented at the Eleventh Annual Conference of the International Association of Chinese Linguistics, Nagoya, Japan. Matthews, Stephen & Yeung, Lousia Y.Y. 2001. Processing Motivations for Topicalization in Cantonese. In K.Horie & S.Sato (eds), Cognitive-functional Linguistics in an East Asian Context, pp. 81-102. Tokyo: Kurosio. Matthews, Stephen & Virginia Yip. 1994. Cantonese: A Comprehensive Grammar. London: Routledge. Morey, Stephen. 2006. Nominalisation and Its Discontents: Clausal Nominalisation in Nimhpuk Singhpo. Paper presented at the Workshop on Nominalization in TibetoBurman Languages, La Trobe University, Melbourne, August 3. Nishi, Yumiko. (2006). The Emergence of the Complementizer no in Japanese Revisited: The Dual Path Approach. Japanese/Korean Linguistics, volume 14. Noonan, Michael. 1997. Versatile Nominalizations. In Joan Bybee, John Haiman and Sandra Thompson (eds.) Essays on Language Function and Language Type. In Honor of T. Givon. Amsterdam: John Benjamins. Noonan, Michael. To appear. Nominalization in Bodic Languages. In Maria Jose LopezCouso and Elena Seoane (eds.), New Reflections in Grammaticalization 3. Amsterdam: John Benjamins. Park, Joo Young. 1999. A Study of Grammaticalization of ‘kut’. MA thesis, Sang Myung University, Seoul, Korea. [In Korean]. (cited in Rhee, this volume). Rhee, Seongha. In press. On the Rise and Fall of Korean Nominalizers. In Maria Jose LopezCouso and Elena Seoane (eds.), New Reflections in Grammaticalization 3. Amsterdam: John Benjamins. Kamus Dewan, new edition. (1989). Sheikh Othman bin Sheikh Salim (Chief Editor). Kuala Lumpur: Dewan Bahasa dan Pustaka.


Shi, Yuzhi & Charles N. Li. 2002. The Establishment of the Classifier System and the Grammaticalization of the Morphosyntactic Particle de in Chinese. Language Sciences, 24, 1-15. Shin, Mikyong. 2005. Development of the Adnominal System in Some East Asian Languages. MPhil Thesis, Chinese University of Hong Kong. Shinzato, Rumiko. 2005. Nominalization in Okinawan: A Diachronic and Comparative Perspective. Manuscript, Georgia Institute of Technology. Sio, Ut-Seong Joanna. 2007. The adnominal and sentence-final ge3 in Cantonese. Manuscript, Department of Chinese and Bilingual Studies, Hong Kong Polytechnic University. Simpson, Andrew. To appear. The Grammaticalization of Nominalizers in Burmese. In Maria Jose Lopez-Couso and Elena Seoane (eds.), New Reflections in Grammaticalization 3. Amsterdam: John Benjamins. Simpson, Andrew. 2003. On the Reanalysis of Nominalizers in Chinese, Japanese and Korean. In Audrey Y.H. Li & Andrew Simpson (eds.), Functional Structures, Form and Interpretation: Perspectives from East Asian Languages. London: Curzon Routledge. Simpson, Andrew & Xiu-Zhi Zoe Wu. 2001. The Grammaticalization of Formal Nouns and Nominalizers in Chinese, Japanese and Korean. In T. E. McAuley (ed.), Language Change in East Asia, 250-283. Richmond, Surrey: Curzon. Simpson, Andrew & Xiu-Zhi Zoe Wu. 2002a. IP-raising, Tone Sandhi and the Creation of Sentence-Final Particles. Journal of East Asian Linguistics, 11(1), 67-99. Simpson, Andrew & Xiu-Zhi Zoe Wu. 2002b. From D-to-T: Determiner Incorporation and the Creation of Tense. Journal of East Asian Linguistics, 11(2), 169-209. Stassen, Leon. 1997. Intransitive Predication. Oxford: Oxford University Press. Sweetser, Eve. 1996. Mental Spaces and the Grammar of Conditional Constructions. In Gilles Fauconnier & Eve Sweetser (eds.), Spaces, Worlds, and Grammar. Chicago: Chicago University Press. Tabor, Whitney & Elizabeth C. Traugott. 1999. Structural Scope Expansion and Grammaticalization. In Anna Giacalone Ramat & Paul J. Hopper (eds.), The Limits of Grammalicalization, pp. 227-270. Amsterdam: John Benjamins. Taylor, John R. 1996. Possessives in English: An Exploration in Cognitive Grammar. New York: Oxford University Press. Traugott, Elizabeth C. 1988. Pragmatic Strengthening and Grammaticalization. Proceedings of the Fourteenth Annual Meeting of the Berkeley Linguistic Society, pp. 406-416. Traugott, Elizabeth C. 1995. Subjectification in Grammaticalization. In Dieter Stein & Susan Wright (eds.), Subjectivity and Subjectivisation, pp. 37-54. Cambridge: Cambridge University Press. Traugott, Elizabeth C. 2003. From Subjectification to Intersubjectification. In Raymond Hickey (ed.), Motives for Language Change, pp. 124-139. Cambridge, UK: Cambridge University Press. Traugott, Elizabeth C. In press. From Ideational to Interpersonal: A Reassessment. In Hubert Cuyckens, Kristin Davidse & Lieven Vandelanotte (eds.), Subjectification, Intersubjectification and Grammaticalization. Berlin: Mouton de Gruyter. Traugott, Elizabeth C. & Ekkehard Konig. 1991. The Semantics-Pragmatics of Grammaticalization Revisited. In Elizabeth C. Traugott & Bernd Heine (eds.), Approaches to Grammaticalization, vol. 1, pp. 189-218. Amsterdam: John Benjamins.


van Breugel, Seino. 2006. Nominalisation in Atong. Paper presented at the Workshop on Nominalization in Tibeto-Burman Languages, La Trobe University, Melbourne, November 2. van Minde, D. 2003. ―More Notes on Malay Yang.‖ Paper read at the Seventh International Symposium for Malay and Indonesian Linguistics (ISMIL), Nijmegen, The Netherlands, June 26 – 29. [The Pragmatic Function of Malay Yang.] To appear in Proceedings of the 7th International Symposium of Indonesian and Malay Linguistics (ISMIL). Wang, Li. 1980. Hanyu shigao. Beijing: Zhonghua Shuzu. Watters, David E. To appear. Nominalization in Himalayish Languages. In An Encyclopedia of Nepal‘s Languages. Wrona, J. 2005. Specificational Pseudoclefts in Old Japanese. Folia Linguistica Historica volume 26. Wrona, Janick. 2005. The Adnominal Form in Old Japanese: Grammaticalization and Degrammaticalization of an Inflected Form? Paper presented at the Workshop on The Development of Nominalizers: East Asian Perspectives, Third International Conference on New Reflections in Grammaticalization (NRG3), University of Santiago del Compostela, Spain, July 17-20 Yap, Foong Ha & Stephen Matthews. To appear. The Development of Nominalizers in East Asian Languages. In Maria Jose Lopez-Couso and Elena Seoane (eds.), New Reflections in Grammaticalization 3. Amsterdam: John Benjamins. Yap, Foong Ha, Stephen Matthews, & Karou Horie. (2004). From pronominalizer to pragmatic marker: Implications for unidirectionality from a crosslinguistic perspective. In Olga Fischer, Muriel Norde & Harry Perridon (Eds.), Up and Down the Cline: The Nature of Grammaticalization, pp. 137-168. Amsterdam: John Benjamins. Yip, Virginia & Stephen Matthews. 2007. The Bilingual Child: Early Development and Language Contact. Cambridge, UK: Cambridge University Press.


Gil also identifies –nya as another variant of punya, but this link appears controversial. Englebretson (2003), in an analysis of colloquial Indonesian, identifies epistemic (stance) functions of –nya with pronominal and nominalizer –nya instead. Less controversial is the reduced form pun, which has been identified as a genitive marker in Ambonese Malay (Holm 1989; Ansaldo & Matthews 1999:45). 2 Tokens of empunya and its phonologically reduced variant punya for this study come from classical texts (prose only) from the Malay Concordance Project (courtesy of Dr. Ian Proudfoot and his research team at the Australian National University; The originals of some of these texts date as early as the 14 th century; the latest ones were written in the 19th century. In analyzing the classical texts, I have used the textual chronology provided in the Malay Concordance Project. The earliest available versions date from around 1600. Some of the contemporary colloquial examples of punya and mya come from the field notes of David Gil (1999), gathered from kampungs ‗villages‘ in the Kuala Lumpur vicinity in the 1990s. Other examples are based on similar Malay varieties, spoken in south and central Perak (a west coast state in West Malaysia). 3 Uses of empunya dia and empunyanya, both meaning ‗its master/owner‘ and the latter highlighting a second layer of –nya encliticization as in () below, suggests that in the mind of the native speaker lexical noun empunya is already typically construed as a single unitary lexeme rather than two dissociable morphemes (empu + -nya). () ―Hei Kastam, tiada pula aku tau siapa empunya -nya.‖ INTJ (name) NEG even 1SG know who master 3SG ‗Hey Kastam, I don‘t even know who its master/owner is.‘ (AH 105:3, 16 th century)



Although the canonical (i.e. non-empunya type) possessive does not have a pronominal counterpart, equivalent effects of possessor vs. possessee focus can be achieved by stressing one NP relative to the other. For example, a stress on the first NP of emas aku highlights the possession emas ‗gold‘, yielding the interpretation ‗my gold‘ (as opposed to some other possession such as pearls, or land, or some other item), while a stress on the second NP highlights the possessor aku ‗I‘, yielding the interpretation ‗my gold‘ (as opposed to someone else‘s). The latter interpretation is functionally equivalent to the headless pronominal punya, as in aku punya ‗mine‘. The headed possessive punya construction, as in aku punya emas ‗my gold‘ is neutral in terms of NP focus. 5 Feature spreading is not uncommon. Visible evidence of feature spreading is reported in research on sign languages within the domains of negation, topicalization and interrogatives (e.g. Aarons 1994; Coerts 1992; Lee 2005; Liddell 1980; Petronio & Lillo-Martin 1997). 6 Note that enclitic –nya serves a complementizing function in negative polarity cleft constructions involving Bukannya, translatable into English as ‗It‘s not that...‘ This explains why the appearance of an overt head-initial complementizer such as yang (equivalent to English that) is unnecessary. See also Englebretson (2003) for arguments that enclitic –nya has not been reanalzyed as a complementizer but nevertheless contributes to complementation strategies. 7 A lexical noun interpretation (hence non-pronominal reading) for empunya is also possible, as illustrated in ()—in this case, empunya highlights the possessor NP. The difference between the lexical and pronominal reading hinges largely on degree of emphasis (i.e. force) placed on empunya, as seen in the more forceful cleftlike reading of (b), relative to the more neutral reading of (a). () Bukannya hamba empunya Ø NEG.3SG 1SG.HUMBLE own(er)/GEN a. pronominal reading: ‗It‘s not mine.‘ b. non-pronominal reading: ‗It‘s not that I (am) the owner.‘ a‘ Bukannya [NPpossessor hamba] [GEN-Pronominal empunya] [NPpossessee Ø] b‘ Bukannya [NP hamba] [V zero-copula] [NPpossessor empunya] [NPpossessee Ø] 8 Not all head nouns are readily elidable. As seen in () below, some abstract referents are not that easily retrieved when elided. () Ini anak saya punya pekerjaan/*ø. this child 1SG GEN NMZ-work-NMZ ‗This is my child‘s doing.‘ (admission of one‘s child‘s mischief) 9 Among the most productive tin-mining states are Perak and Selangor. These states have a high concentration of Chinese immigrants, particularly in urban areas related to the tin-mining industry. Settlements along the Straits of Malacca also attracted many Chinese imimmigrants, many of whom also engaged in trade. These settlements included the flourishing ports of Penang, Malacca and Singapore. Malacca, in particular, already had a small but well-established Chinese community from the days of Princess Hang Li Po, who was brought as a princess bride to the royal courts of Malacca in the 15 th century by Admiral Cheng Ho. This took place during the Ming Dynasty, when China reached its peak as a maritime power. 10 It is possible that some of the contact phenomena related to the grammaticalization of empunya had occurred earlier in Bazaar (i.e. ‗market variety‘) Malay and Baba Malay. These language contact varieties developed from close trading contacts with China in the 15th century, and possibly even earlier (Moser 1985; Ansaldo & Matthews 1999). The influence from Bazaar Malay and Baba Malay on the emergence of genitive punya, however, was not found in the pre-19th century Malay classical texts that we examined. This may have been due to register differences, with Bazaar Malay and Baba Malay being restricted to colloquial usage, not appearing in printed texts until later. 11 During this time Malay replaced English as the primary medium of instruction in public schools in Malaysia. 12 Lee (2006) recently observed that [AdjP] ge [N] constructions do occur in Cantonese as marked constructions, i.e. very low in frequency and possibly sometimes with focus interpretations. This appears to be a recent development.Whether this phenomenon is due to reverse contact-induced grammaticalization among the Chinese diaspora in Southeast Asia which found its way back to southern China is as yet unclear. 13 These constructions are not unlike the punya relative clauses in Bazaar Malay, as discussed in (29b) earlier and in () and () below.


Saya tadi guna punya pinggan belum cuci (lagi). 1SG use REL plate not-yet clean yet ‗The plate I used just now has not been washed (yet).‘ () Saya tadi guna punya belum cuci (lagi). 1SG use PRONOMINAL not-yet clean yet ‗The plate I used just now has not been washed (yet).‘ 14 While intensifier punya can readily modify most adjectives (e.g. punya tinggi ‗so tall‘, punya biru ‗so blue‘, punya bersih ‗so clean‘, etc.), only a restricted set of verbs (e.g. attitudinal verbs) can be modified by intensifier punya (e.g. punya minat ‗really admire‘, punya benci ‗really hate‘, punya takut ‗really fearful‘). 15 For more detailed discussion of sentence-final particles in Cantonese, see Matthews and Yip (1994) and Chor (2004). 16 Interestingly, corpus analysis of natural (i.e. non-elicited) spoken Cantonese also shows that prenominal prepositional phrases (PPs) decrease in frequency—being replaced instead by topicalized constructions—as word length of the modifying phrases increases (Kwan 2004, 2005). 17 In this regard it is worth bearing in mind evidence from typological (Greenberg 1966; Dryer 1992) and cognitive processing (Hawkins 1990, 1994, 2004; Kwan 2004, 2005; Matthews & Yeung 2001) literature, which highlights that prenominal modification is cognitively costly for SVO languages. It is thus not surprising that these clause-internal stance expressions tend to involve very short phrases (e.g. punya + ADJ in Malay; gam (*ge) + ADJ in Cantonese). 18 Note that sentences (48) to (50) are reproduced verbatim from the internet; I have chosen to retain the colloquial flavor and thus have not standardized the spelling. 19 Code-mixing/code-switching is quite common in colloquial registers in contemporary Malay. 20 More neutral cause/reason constructions in Malay employ connectives such as sebab, kerana and pasal (or fasal)—all equivalent to ‗because‘ in English. As seen in () and () below, these connectives can appear in either clause-initial or clause-final position, unlike punya pasal, which appears only in clause-final position. () Sebab/Kerana dia cemburu, kekasihnya jadi mangsa. () Kekasihnya jadi mangsa sebab/kerana/pasal dia cemburu. 21 ‗Say‘ verbs are frequently recruited as complementizers in Chinese and other languages. Examples include Taiwanese kong (Simpson and Wu 2002a) and numerous examples from the Kwa languages of West Africa (Lord 1993). 22 In this regard, it is worth noting that Cantonese XP ge3 jyun4jan1 (‗the reason for XP) constructions are similar to uses of Malay XP punya pasal constructions where pasal still retains head noun status and a lexical noun reading (i.e. ‗matter (as in affairs), reason, cause, fault, etc.‘). In this regard, the Cantonese XP ge3 jyun4jan1 constructions are headed relative clause constructions. Decategorization of head noun jyun4jan1 (‗reason‘) to a formal noun has not taken place, hence the emergence of a causal subordinator in the form of Cantonese ge3 jyun4jan1 has not been facilitated. One can argue in the case of Cantonese conditional subordinator ge3waa2 that waa6 ‗word(s)/speech‘ has undergone decategorization to a formal noun waa2, which then combines (or ‗merges‘) with ge3 to form a complementizer that serves a conditional subordinator function.